Compositional data analysis in time-use epidemiology

Abstract

How we allocate time to activities impacts our health. Daily times spent in activities are interrelated because they compete for time-shares within a finite 24 h window. If more time is spent in one activity, time must be taken from one or more of the remaining activities to maintain the fixed total of 24 h. Thus, time-use data have a relative nature and can be analysed accordingly using compositional data analysis. In this chapter, we demonstrate exploratory and cross-sectional inferential analyses of an eight-part time-use composition using data from the Longitudinal Study of Australian Children (.n. = 2224, 52% boys, mean age = 34 months, standard deviation = 3). For inferential analyses, time-use compositions are expressed as a specific choice of balance coordinates to separate between types of activities. Considering the balance coordinates as explanatory variables, we explore the relationship between children's time-use composition and their socio-emotional health. Subsequently, we consider the balance coordinates as dependent variables and explore the relationship between parental perception of neighbourhood liveability and their child's time-use composition.